102 



Mr. J. C. McConnel on the 



when tilted to the left. In the first determination I used sky 

 light, but, finding that on Fresnel's theory the depolarizing 

 power of the pile of plates was almost independent of the 

 colour of the light, I decided in the second determination to 

 avail myself of the much greater brightness of sunlight reflected 

 from a sheet of white paper. 



Aug. 13, 1888.— Sky hght, Davos. 



G 



0°. 



30°. 



40°. 



50°. 





d) 



37°7 

 39°-3 



34°-7 

 35° 



30°-3 

 31°-2 



27° 

 25° -5 



(b 





Sept. 14 and 15, 1888.— White light. 



G 



40°. 



50°. 



60°. 





d) 



30°-l 

 29°-9 



26° '8 

 26°- 1 



22°-7 

 21°-0 









These various readings are not as accordant as I should 

 have wished. The probable error of the figures given is nearly 

 twice as great as that of single observations of the sky. This 

 increased uncertainty arose from the smallness of the field of 

 view in the standardizer. The aperture of the polarizing Nicol 

 was about 15 millim. by 10 millim., and its distance from the 

 eye about 33 centim. On the other hand, in taking a reading 

 at a high angle of incidence, such as G=60°, the standardizer 

 had an advantage in that there was no difficulty in getting a 

 complete reversal, changing the black cross into white, or vice 

 versa. Whereas in taking an observation of a strongly pola- 

 rized sky, the plates had to be moved, and this difficulty always 

 occurred. For this reason observations of the sky for which G 

 approaches or exceeds 60° are somewhat untrustworthy. 



It will be noticed that I had to follow Rubenson^s example 

 in using a much smaller aperture for standardizing than for 

 observing. I used a pile of five plates only * while Rubenson 



* My pile of five plates was about 10 millim. thick, which is less than 

 the apertm-e of the Nicol in the direction which was of importance ; Ru- 

 benson's pile was apparently more than twice as thick as his aperture 

 was wide. 



